bogus bogotuning
Boris 'pi' Piwinger
3.14 at logic.univie.ac.at
Wed Jan 28 14:27:20 CET 2004
Jason A. Smith wrote:
> I have already had to tweak the default parameters in order to achieve
> better results, but I am only able to guess at numbers since you won't
> allow me to run bogotune to even just experiment a little.
Here is an alternative I can offer to you: If you do full
training you need to find the right parameters. If OTOH you
do pure train-on-error, the training adjusts to your
parameters, they are built into the database in some sense.
It seems the parameters change the size of the database, but
I have not seen evidence that they change the precision. The
thing you have to do beforehand is setting your spam_cutoff
to some value in the middle. See
http://cvs.sourceforge.net/viewcvs.py/*checkout*/bogofilter/bogofilter/doc/bogofilter-faq.html?rev=HEAD&content-type=text/html#training
(Method 2) for details.
Maybe this method works for you even with few messages.
Would be interesting to know.
I personally use these settings:
http://piology.org/bogofilter/.bogofilter.cf
pi
More information about the Bogofilter
mailing list